...Starting with first principles
America First Trust services
Education, analysis, and advice for uncertain times

Mining and Robotics:
An Introduction

A next new thing joins the commodities bull market

by Bill Fox

Background for the December 10, 2004 presentation to the New Technologies in Exploration & Operations Session of the 110th Annual Meeting and Exposition of the Northwest Mining Association. Updated Feb 22, 2005.



I have a dream that some day planetary rover-type robots will systematically and tirelessly collect and analyze soil samples for junior mining explorationists on earth...that armies of snakebots will autonomously go down bore holes and continuously worm their way through vast underground strata and produce assays and 3-D maps of suspected ore bodies at almost any depth...and that submersible mining robots connected to long continuous process ore tubes will mine around the clock while chasing veins to super hot earth depths unimaginable to human miners today. I also dream that certain mines will become totally automated, and that mining executives will be able to pull up on their lap tops reports that summarize everything about their mining operations, to include 3-D rotating maps and complete financial, geological, and production analyses.

Intuitively, I think that if mining companies were to aggressively pursue any of these concept areas, that within five to ten years they could produce next generation exploration and mining robots that could perform current tasks with ten times the efficiency and less than ten times the current costs. And these improvements would barely scratch the surface of even greater next generation advances that would lie ahead. Last, but not least, by pursuing visionary projects, I believe that mining companies can not only help themselves, but can also help to advance major national strategic priorities.

But I get ahead of myself. First we need to address the here and now. I say this, believing that the fledgling steps being taken by mining companies and roboticists today will no doubt appear very quaint to miners fifty years from now, comparable perhaps to the way horseless carriage contraptions at the turn of the century look to us today.

In this paper, I will address a number of themes about the mobile robotics industry identified in my overview "I, Robot Investor," as well as examine variations and paradoxes that apply to the mining industry. This is intended as a conceptual introduction to the overall field rather than as a profile of any publicly traded companies that might make good investments today, indeed, the current players in the mining-related robotics area are typically very small privately held companies or else, in regard to publicly traded firms, subcomponents of a technical services department (as in the case of Placer Dome of Vancouver, B.C.) or else wholly owned subsidiaries (as in the case of MD Robotics relative to MacDonald Dettwiler of Richmond, B.C.) that are still way too early in the R&D cycle to have any kind of significant or even visible impact on overall corporate profitability.

The view at the top

On a macro level, across all industries, we see mobile robotics invading all areas of the structure of production, ranging from raw materials to finished goods. So-called "Moore's Law," postulated by Gordon Moore in 1965, will likely provide a powerful wind to the back of robotics for decades to come. In its current incarnation, Moore's Law predicts a doubling of processing power every 20 months or so from steadily increasing transistor densities on chips, increasing clock speeds, stacking concentrations, and other factors. It may take another twenty years to achieve on a chip the processing power of roughly 100 trillion instructions per second of the human brain.

A large near term constraint for robotics in general, and mining robotics in particular, involves navigation and manipulation. A large portion of human brain capacity is devoted to storing and processing vast libraries of 3-D images required to interpret and manipulate the physical world. Robots currently function on the approximate perceptual level of two month old human infants. Their intelligence is also often compared to that of insects.

[Left]: Projected worldwide mobile robotics growth by the Japanese Robotics Association, reproduced from Sept 18, 2003 Dan Kara article:“Analysis – Sports: Sizing and Seizing the Robotics Opportunity.

Another constraint involves the need for certain robotic modules to increase in capabilities and simplicity of use and to come down in price to the point that the benefits of their particular application becomes economically obvious. The accompanying chart by the Japanese Robotics Association shows a prediction made by many technology forecasters. They believe that price points are already hitting levels that should encourage accelerating mobile robotics applications across industries.

Despite the exponential growth of the "Moore's Law" trend, attempts to "robotize" various parts of the mining industry in the last decade have often been intermittent affairs characterized by the proverbial two steps forward, one step back, two steps forward, and one step back. This is often due to many factors outside the scope of management. Certain factors include the complex and varied nature of mining operations. Another important factor involves the low commodity prices in the late 1990's that put a freeze on mineral exploration, not to mention funding for advanced technology projects.

According to Dr. John P.H. Steele at the Colorado School of Mines, we can have robots doing some tasks in certain types of mines within five years. For a mine to be "robotized," we are talking about a complete makeover. Paul Semple, President of the Canadian robotic systems integrator Penguin ASI, thinks current semi-autonomous robot system technology can potentially save certain underground mines about 30% of their current operating costs, and open pit mines 15% of their costs. He emphasized that these numbers are by no means fixed in stone, since the production and cost characteristics behind each mine are different.

It is instructive to step outside the mining industry and look at the kind of investment that is going into mobile robotics elsewhere. In the consumer products area, major Japanese firms such as Sony, Honda, and Toyota are pursuing visions of humanoid robots such as the QRIO and Asimo as affordable domestic play toys and servants in five to ten years. They are collectively spending the equivalent of tens of millions of U.S. dollars a year on R&D in a "brute force" effort to get there, regardless of current market economics.

In July 2003, neuroscientist Dr. Mitsuo Kawato, Director of ATR CNS Laboratories in Japan, called on the Japanese government to commit $446 million a year for the next three decades for the "Atom Project." He wants to build a robot with the mental, physical, and emotional capabilities of a five year old child, and mimic inside the robot all the different functional areas of the human brain.

Dr. Kawato is probably referring to Atom Boy, a robotic cartoon character popular in Japan since the 1950's. While attending the Carnegie Mellon Robotics Institute 25th Anniversary Event in October 2004, I saw Dr. Kawato play a video of a robot he has built in cooperation with Sarcos, a Salt Lake City-based U.S. robotics firm. The robot had the autonomous ability to play a drum and bounce a ball on a racket. Dr. Kawato showed us a video featuring a Japanese TV reporter having fun playing a scrabble game against the robot with Georgia Tech researcher Dr. Chris Atkeson standing between the contestants. The Japanese have been very cooperative with American research institutions, and I admire their far-sightedness and disciplined efforts.

Unfortunately American firms are notorious for taking a short term outlook and are heavily indebted. Contrary to glowing outlooks often reported in the national media, I am more inclined to believe data presented in the Grandfather Economic Reports that suggest a long term trend of declining productivity and competitiveness in America.

Imagine if Americans were to one day wake up and find that countries outside their own have created over a million robots with human level intelligence, that can in turn design and produce millions of even more intelligent robots at an escalating rate, and that we have failed to keep up with them. This raises a question about whether the mining industry in America might be able to play an important role in pulling along strategic robotic research and development in an era of rising commodities prices.

Despite the visionary appeal of robotics, economically adapting automation to mining still involves the "old fashioned" process of finding ways to free up bottle necks in the "process flow" that typically starts with blasting rock and mucking (loading) raw ore underground and ends with delivering refined product in an above ground facility. Management still has the duty to thoroughly understand in detail all the steps involved in production and all the tasks performed by workers in order to intelligently apply new automation.

On a macro-economic level, I expect the commodities bull market which began in 2000 will continue for at least another five to ten years. In fact, commodities bull markets have historically been very long, and this could go on for fifteen to twenty years, in essence mirroring the very long U.S. stock bull market that lasted from approximately 1980 to 2000. I wrote an article posted at numerous gold-related Internet sites titled "Back of the Envelope Analysis for $1,000 Gold in Five Years" that provides more details on the factors at play here. (Actually I personally think it is likely we will go well above $1,000 in less time than this, but wanted to be "conservative"). I would also refer the reader to James Puplava's outstanding "Perfect Storm Update" series for more background. The net effect is that extractive industry producers will likely become flush with cash like never before.

Continuing with paradoxes, when higher commodities prices will put the extractive industries in a position where they can most afford to invest in robotic R&D, they will be most tempted to squander their treasure through various forms of "financial" engineering (the 2001 Enron scandal serving as an extreme American example) as opposed to investing in "real" engineering applications that pay very real long term productivity dividends. Conversely, looking back at the very depressed commodities prices of the late 1990's, when natural resource companies needed advanced automation the most to reduce their costs to rock bottom, they could least afford it.

If mining companies squander this coming golden opportunity to fully automate, my guess is that during the next commodities bear market, most US mining companies will get taken over by foreigners who exercise better social discipline and technological far-sightedness. The recent Chinese attempt to take over Canada's Noranda was a dress rehearsal for more of what may come later.

Does the mining industry have favorable robotic implementation characteristics?

On the surface, the mining industry seems to have major characteristics that favor installing robots. First, it has the"3-D's" (work that is dirty, dangerous, and duplicative [or drone-repetitive]). Secondly, the cost of robot systems may be only a small fraction of the overall costs of various types of equipment currently in use and the marginal revenue of the additional long term throughput that they can help produce.

There are also characteristics of the mining industry that create impediments. Robotics installation is still performed on a case by case basis. Each mine has different production and process flow characteristics. Instead of furnishing a mass market with "killer apps," at this stage of robotic technological advancement the mining industry seems to require customized systems integration and specialty contract engineering. However, this should ultimately change as mobile robots become increasing intelligent and versatile.

Characterizing the mining "robotization" process

First, let's define "robotization." The focus of this paper is on mobile robotics, which typically means the last foreseeable phase of automation where machines can move themselves and can sense and react to their environment with varying levels of intelligence. The term "robotization" might also include fixed robots, which have been around on automobile assembly lines since the 1970's. Typically a fixed robot is a machine capable of behavioral sequences, with a very limited ability to sense the environment. It's base is usually fixed to a specific location.

Automation phases. As some background to "robotization," it is helpful to see how it fits into the sequence of overall automation steps projected for the mining industry, as depicted in the study below completed in 1992. In actuality, only the first rapid communications phase has been implemented. All the other phases have lagged behind projections.




Evolution to Automation, a study carried out in 1992 by Hatch Associates in Canada for Industry Science & Technology of the Canadian Federal Government. Courtesy of MD Robotics


The last automation phase depicted among the four charts shown above is labeled as "autonomous mining."

What does "autonomous" mean? Currently some robotic systems integration web sites for the mining industry use the term "autonomous" when what they really mean is "teleoperated."

To get even more specific, "teleoperated" typically means taking a basic mining vehicle originally designed for a human driver such as a Load Haul Dump (LHD) vehicle, and adding remote control equipment.

An example of a remote control-equipped Load-Haul-Dump vehicle is provided below. The function of an LHD vehicle in a typical underground mine is to scoop up (or "muck") ore from where it is blasted away from a rock face, carry (or "tram") it to an elevator site, and then dump the ore on the elevator platform for transport to a surface processing facility.

A Load-Haul-Dump (LHD) vehicle adapted for robotic teleoperation. Source: MD Robotics.


"Teleoperation" involves ways that humans can control robots in real time over the Internet, by wireless, or by some other communication medium. It typically requires high band width to process video signals sent from robot cameras to human controllers. Teleoperation also involves controlling robot movements. This can include joysticks and foot pedals. It can also include an exoskeleton that fits over human hands that controls the movement of robot hands.

To be truly and completely autonomous, machines would need to be able to negotiate their environment on their own and perform complex tasks without human supervision. In his book Flesh and Machines, Dr. Rodney Brooks of the MIT Artificial Intelligence Lab thinks that robots will require human teleoperation supervision to perform complex navigation and manipulation tasks for at least the next ten years.

Scientists have already produced a supercomputer with a third the capacity of the human brain. If scientists link enough computer chips together, they can eventually create almost as much processing power as they want, although this obviously gets into storage space, running speed, and software integration considerations. They can also get around the problems involved in trying to fit all of this processing power inside a single mobile robot by creating wireless connections. However, major economic constraints involve such factors as the unwieldiness of these new visionary systems and the costs of developing software designed for special applications of processing power that are way ahead of the current microprocessor generations.

How do you predict how long it will take research teams to solve hardware and software problems involved in specific robotic applications and ultimately mimic all aspects human intelligence? Intuitively it seems like economically pulling all this together into a single mobile robot will require at least another five to ten computer chip development generations. In the meantime humans will need to use teleoperation to supplement and override machine operations for complex tasks.


Teleoperator control of two Load-Haul-Dump (LHD) vehicles on different levels. This and preceding pictures courtesy of Automated Mining Systems/MD Robotics.


A "robotization" program aims at a lot more than simply taking a driver in a mine out of a Load-Haul-Dump (LHD) vehicle, moving him to a teleoperation center on the surface, and then asking him to maneuver an LHD vehicle all day with a joystick and foot pedals. To go beyond "remote control" to "robotization" we use artificial intelligence programs that remember all the instructions to the LHD vehicles under varying conditions. The artificial intelligence programs can then enable LHD vehicles to increasingly perform certain behaviors on their own. This can include such behaviors as dumping ore at certain sites, or remembering how to go about tramming ore between the loading and dumping sites on their own.

As he successfully "shapes" the behaviors of different vehicles, the teleoperator may come to control six or more vehicles simultaneously. His operations have to be coordinated with other systems involved in mine operations. Obviously the teleoperator is more than an equipment operator. He may also be involved in various levels of robot programming and systems integration. His ability to proactively help advance a technological strategic plan set down by management may become as important as meeting physical mining production targets.

"Robotization" is occurring everywhere up and down the value chain. A subtle example consists of "drowsy driver" software created by Assistware to monitor for un-signaled drift out of car lanes by heavy truck operators suffering from fatigue. Another example consists of the way Newmont Mining, the largest gold producer in the world, has led the industry by completely automating its assay labs in Carlin, Nevada. Newmont has reduced the turn-around time required to determine the precious metals content of core samples, which in turn speeds up the ability of its open pit operations to distinguish higher grade ore rock from waste rock.

Sometimes a marginal improvement in the performance of a dated machine concept can make a big difference. The May 1998 article, "Australian Scientists Develop the World's Largest Robot," announced that roboticists had inserted robotic controls inside a 3,500 tonne huge walking crane with a 100 meter boon. The crane's drag line reflected 1950's technology. The new controls "increased productivity of a drag line by around 4 per cent, which may not sound all that much - would save the typical Australian coal mine $3 million a year, or $280 million for Australia as a whole." This innovation is currently being marketed in the U.S. by P&H Mining Equipment as the Universal Dragline System.

"Robotization" is embedded in equipment. According to an MD Robotics spokesperson, "A typical LHD might be outfitted with as many as 150 sensors of one type or another. These include sensors to measure hydraulic or engine pressure, air pressure sensors on tires, and accelerometers to sense rocks lying in the vehicle's path." On LHD vehicles at Inco's Stobie Mine, they include fire sensors that alert teleoperators to activate fire suppression systems.

Embedded :"sensor suites" become increasingly "robotic" as sensors activate corrective mechanical action without human intervention. At some point in the future we might imagine intelligent robotic arms that can unfold from vehicles and perform various maintenance functions on their own.

"Robotization" can ultimately mean a complete reconceptualization of mining jobs and operations. Most of our concepts of "work" come from the perspective of a hands-on human operator. However, robots do not require a "human" environment, human capabilities, or a human operational tempo.

In Part Five of my paper "I, Robot Investor," I discuss why replacing human workers per se is typically the wrong focus as well as questionable social policy. The better focus is on empowering humans to accomplish more through advanced automation. It is also better to reconceptualize work tasks outside of exiting paradigms in order to accomplish them in radically better ways.


A robot is really the sum of the modules that make it up. Comparing robot parts with human anatomy is very useful way to introduce the reader to basic concepts. As indicated elsewhere in this paper, it can also be very self-limiting in terms of inventing versatile machines. In Part One of " I, Robot Investor," I use author Gareth Branwyn's overview of modules. He references human analogs in parentheses:
a) Frame (human skeletal system)
b. Power System (GastroIntestinal System)
c. Actuators (Musculature)
d. Drive Train (Legs and Feet)
e. Controller(s) (Brains/Central Nervous System)
f. Sensors (Five senses: Hearing, Touch, Taste, Smell, Sight)
g. Manipulators/End Effectors (Hands and Fingers)
h. Communication (Speech –optional feature) and i. Outer Shell (Skin –optional feature). In Part Three I provide an general overview of modular frontiers across many industries.

All of these modular areas are steadily coming down in price and increasing in capabilities, but at different rates for each type of module. Typically there is kind of cascading effect, where certain modular capabilities will hang in limbo without being used in industrial robotic applications while waiting for improvements in other modules. Once the "bottleneck" module improves to a certain level, it can then be economically combined with other modules to create a radically new robot concept. Placer Dome's MiniMole project, discussed later, appears to be an example of a happy combination of advancements in rock-cutting technology for the end effector (or work-head), drive train (or propulsion) capabilities, and communication (or teleoperation) modules.

Communication capabilities are obviously a big issue with teleoperation, which requires high bandwidth to transmit video, audio, and command signals. An advantage of underground mining is that teleoperators can use whatever bandwidths they wish without intruding on government-mandated frequency assignments. A disadvantage of working in underground mines is that they often entail very unstable and changing environments where one may need to continually string up and move around antennae wire for communications systems.

Teleoperation can be applied over considerable distances. Inco has run teleoperation demonstrations from Montreal to control Load-Haul-Dump (LHD) vehicles at its Stobie mine about 600 miles away in Sudbury, Canada with a response time of about one tenth of a second According to Dr. Rodney Brooks in his book Flesh and Machines, an MIT study determined that teleoperation usually becomes unwieldy with a delay of half a second or greater. This more commonly becomes a problem with attempts to control devices half way around the world over the Internet.


A "ruggedized" teleoperated drilling system. Source: MD Robotics.


Not surprisingly, mining robotic system usually have to be highly "ruggedized." According to the article "Armchair Mining," "[MD Robotics] builds its own electronic components designed to withstand shocks up to 50 g and vibrations that are worse than the Saturn liftoff, [the President] said. Because underground vehicles encounter air pressure changes while traveling through the mine tunnels, controllers are gas pressurized to keep acidic water from passing through seals."

Incidentally, in the picture above, the multiple images on the teleoperator's screen suggests another important aspect of teleoperation. As previously mentioned, the ultimate goal is to use artificial intelligence to increase the autonomous nature of each machine and increase the number of separate machines that a single operator has under his simultaneous control.

Light rope system. Source: MD Robotics

Underground mines lack the benefit of global geo-positioning systems. Their dimensions may be changing as walls are being excavated, ceilings crumble, and heavy vehicles are constantly bumping into things. It is nice to have a navigation system that can work off ambient light and is independent of long light cords on ceilings or sensors implanted in walls. MD Robotics developed its IGS or "Infrastructure-less Guidance System" in which a vehicle "learns" its route as it is navigated by the teleoperator. It uses a combination of dead reckoning and a limited artificial intelligence capability to memorize the length of various routes and the characteristics of intersections. An algorithm keeps it centered in corridors.

As mentioned earlier, computer object recognition capabilities are often compared to those of a two month old human infant. Computers can recognize frontal views of faces, but not age progression. At the October 2004 RoboNexus robot conference, Dr. Rodney Brooks, head of the MIT Computer Science and Artificial Intelligence Lab, showed a video of a robot passing from a hallway into a coffee lounge. Using what he called a "brute force" algorithms that compared total fields of pixels, the robot was able to distinguish the hallway from the coffee lounge. His experimental robot Cog at the MIT lab can follow people with its camera eyes. However, it still has problems manipulating objects such as a slinky coil toy. Software programs for home robots, such as vSLAM developed by Evolution Robotics for its eVac vacuum cleaner, rely on very rough and abstract outlines of walls and furniture.

Dr. John P.H. Steele at the Colorado School of Mines is focusing considerable research on what he calls the last piece of the autonomous LHD vehicle loop. At LKAB's Kiruna mine in Sweden, vehicles can drive themselves through drifts (tunnels) and offload ore by themselves. The problem is in the loading phase. One of Dr. Steele's former students, Paul Lever, wrote a program for scooping that was later acquired by Caterpillar called AutoDig. This recorded loading operations during different cycles and built a computer model for optimum loading for each type of ore.

Caterpillar has apparently put AutoDig on the back burner because it seems to be too advanced for market acceptance right now. According to Woof, "[SIAMtec's] automated loading feature is of particular interest, as it is very different from the Caterpillar Elphinstone system. The SIAMload system is designed to react in real-time (and quickly) using feedback from loads exerted on the hydraulics and structure of the machine. This uses hydraulic pressures, cylinder extensions, axle loads and wheel positions to calculate what's in the bucket - the system does not use software to model the muck piles or try to gauge what's in the pile ahead of each loading cycle. According to Atlas Copco, its system is able to respond quickly and can cope well with over-sized material buried in the pile."

Dr. Steele is interested in writing a more comprehensive program that not only instructs the robot vehicle on how to approach a muck pile and get a full bucket load, but also avoid running into walls. He is researching the creation of three dimensional visual models using stereoscopic vision to deal with broader perceptual problems. He feels that visual systems are better than lasers for dealing with motion problems.




MD Robotics has developed Instant Mine Modeler, or iMM (left), which uses a pair of cameras to stereoscopically build a 3-d virtual model (upper right) or a virtual map of the interior of the mine (lower right).

As a miner sweeps a scene with the Instant Mine Modeler camera depicted above, a computer pieces all the images together into a 3-D virtual model, and can determine distances to hundreds of thousands of points. This data can be loaded into Vulcan or Datamine mine management software to create 3-D maps of mine interiors. It allows management to know immediately and exactly how excavation is proceeding.

Coincidentally, research into building stereoscopic 3-D models is also being done for material handling and warehouse applications by Carnegie Mellon professor Hans Moravec. Dr. Moravec intends to commercialize his research as Chief Technical Officer of Seegrid Systems. One of the reasons why he prefers ambient light to lasers is that ambient light systems are less obtrusive and can use cheap cameras.

Laser devices in action.
Source: MD Robotics

In Sept 2002, WMC Resources announced that they, along with an Australian consortium that included Caterpillar Elphinstine, had developed a "smart loader" at their Olympic Dam mine in Australia following four years of research and development. The loader was both teleoperated or co-piloted by on-board laser sensors and computers that prevented the loader from hitting a wall. It built a map of the underground structure as it went along, and compared it to an abstract mine map which controlled the on-board guidance system. It did not rely on external navigation aids such as light guidance [long light tubes fixed to ceilings] or reflective strips.

Coincidentally, in the laser arena, Carnegie Mellon professor Dr. Red Whittaker has also developed a laser mine mapping system that is more accurate than ambient light systems for determining distances. It can also determine the dimensions of drill hole voids. Dr. Whittaker and his brother seek to commercialize this technology through Workhorse Technologies.

If the reader senses a debate over which is better, lasers or ambient light systems, I would offer a third alternative. As costs come down, use both. One of the advantages of robots is that they can become loaded with as many different types of sensors as one can afford. Data from these sensors can be continuously fed into computer data base systems. They can perform variance analyses and utilize other methods to optimize production. David Hyams, Chief Technology Officer of Seattle-based robotics systems integrator Coroware, has observed that a robot is often just the tip of the iceberg of a database management and retrieval system.



The first area involves "robotizing" equipment originally designed for humans in a human-friendly mining environment. The easiest projects have involved the material handling side of mining, to include robotizing Load-Haul-Dump (LHD) vehicles and trucks. The focus has been on using automation to move ore faster and cheaper through existing operational pipelines using conventional equipment and work concepts. This is "first generation" robotics.

The second generation involves specifically designing equipment as teleoperated robots. Placer Dome's MiniMole project and Sandvik Tamrock's ARM 1100 machine apparently fit in this "second generation" category. Here, the development of a new "end effector" (the part of the robot that does the actual work) such as the new rock-boring Oscillating Disc Cutter is often every bit as important as the underlying robotic system that carries it.

The third generation involves developing robots that completely redefine work paradigms and operational concepts. Robots go where no human could possibly go and work in ways that humans can only imagine but not imitate. Submersible robots designed for deliberately flooded mines are one good example.



Four mining equipment companies have used teleoperation technology to make conventional operations and equipment "faster and cheaper." They are Atlas Copco Wagner, Caterpillar Elphinstone, GHH Fahrzeuge, and Sandvik Tamrock. Quite a few mines have participated in robotized vehicle operations around the world, such as LKAB's Kiruna mine in Sweden, Inco's Stobie mine in Ontario, Canada, the Werra Mine in Germany, Noranda's Brunswick mine, and WMC Resources' Olympic Dam mine in Australia. In addition to these existing robotized operations, Codelco's El Teniente copper mine in Chile and Rio Tinto's Palabora underground mine in South Africa have extensive plans in place to add teleoperated automation.

The April 2003 World Mining Equipment article "A Long Haul" by Mike Woof provided an excellent overview of the LHD and trucking automation area. According to Woof::

The key issues for automation are productivity and utilization, though there are other benefits too. Mines stand to gain from automated loaders and trucks as the equipment should be able to operate at higher speeds without risk of hitting the walls and for a full shift without slowing - computers don't get tired. Trials with Sandvik Tamrock's system in the firm's test mine show that automated loaders can negotiate sharp turns time after time without any scrapes and at far higher speeds than even experienced operators could achieve. According to Caterpillar Elphinstone, trials at its joint venture DAS project show autonomous loaders can bring up to 15% greater productivity/shift, as well as reducing overall capital costs and allowing equipment to work in restricted areas. And with the joint Atlas Copco Wagner/Noranda system, automatic loading means buckets are filled with heavier loads and some 24% quicker than by an experienced operators using either tele-remote or line-of-sight remote controls. The Noranda mine has also been able to increase peak haulage capacity from 5,000-5,900 tonnes/day and set a theoretical level of 7,000 tonnes/day in a 15 hour shift.

Automation is good for equipment as it means that operating functions like gearshifts can be optimized while stresses on diesel engines and electric motors are reduced by overload prevention. This lowers fuel/energy consumption and more importantly, minimizes wear and tear. For example, when Swedish iron ore miner LKAB first started using its automated Toro 2500E loaders (which run on an earlier technology), there was a noticeable reduction in tire wear. In the long term, mines should increase productivity/week as well as reducing production halts caused by unscheduled machine downtime, as the equipment runs faster, more efficiently and for longer. As automated loaders can operate in hazardous areas and return to the face immediately after blasting without any health worries over fumes from explosives, safety is improved in this respect also. Well, that's the theory at least.

...In itself, automated haul truck should utilize a similar level of technology to an LHD and offer similar benefits. The Noranda/Atlas Copco Wagner SIAM project includes automated truck haulage, with the machines working in specific areas in the Brunswick mine and using vision navigation to follow light reflecting tape on the ceiling. This set-up is proving its worth, as production output is greater than for directly-operated vehicles. Average loaded tramming speeds have increased from 8-11 kph in the dedicated 400 m drift where the trucks run under auto-guidance and the system could also extend daily equipment operating hours from the present 15 hour/day shifts...

Woof mentions how the Japanese firm Komatsu envisions a surface truck automation project where robo-trucks can be running in the same areas as manned trucks. This means creating a sophisticated dispatch system to monitor truck locations, as well as radar/ultrasonic/infra-red detection systems on the trucks themselves to avoid hold-ups or accidents. Many mines use gates and other safety devices to prevent workers from getting in front of autonomous vehicles, but in some cases it is not practical to segregate autonomous and manned vehicles.

Artist concept of DARPA
Grand Challenge vehicles

Incidentally, the concept of mixing autonomous and manned vehicles is also being pursued outside the mining industry. The U.S. Congress mandated the Army to have as its goal one robot for every third Army vehicle by the year 2015. Improving autonomous vehicle capabilities is also the focus of the ongoing annual DARPA Grand Challenge races in the Mojave Desert.

The value proposition: The report "Is There A Business Case for Underground Automation?" published Dec 10, 2003 by MD Robotics, discusses how automation can keep vehicles productive more hours during a shift. The transit time for workers from the surface down to the work site can create substantial dead time, particularly for deep mines. Disruptions from blasting or frequent maintenance inspections can also be a big problem. According to the report:

Mine production reports continue to show that an 8-hour operating shift turns into 4 1/2 hours of effective production drilling or mucking [loading vehicles with ore] . There are many reasons for this, however, in spite of efforts made to increase this number, it seems to defy significant improvement. Mines have experimented with longer shifts, only to learn that the ratio of hours worked to length of shift did not significantly increase. The exception in favor of 10-12 hour shifts occurs when travel time to and from work faces is excessive. Even in this case however, operators still become fatigued and productivity falls away in any case.

Recent business case studies completed by AMS [Automated Mining Systems, owned by MD Robotics] show that adding automation to the mobile process clearly has a measurable and significant impact. These studies show that the greatest value that automation can add to the mining process, is to drill more holes and to move more tons per shift. If a mine plan can show an increase of 50% more drilled holes, or material moved per shift, paybacks are quite incredible. Properly implementing autonomous drilling or mucking cycles could mine an ore-body in 30% less time, adding millions of dollars to the bottom line.

In one business case, an underground precious-metal mine, remotely tramming muck along two rail haulage levels between shifts has shown a significant increase in material moved. This translated to a bottom line increase of profit over the first 5 years of almost an order of magnitude greater than the investment. The payback for the installation of the autonomous tramming system was just a few months.

In a second business case, an underground (base metal) block cave application, creating an autonomous muck cycle for 3 LHD's showed a 30% improvement in operating time per shift, and a corresponding increase in material moved. The payback was less than 7 months. The value of the extra ore moved over the first 5 years was over 7 times the investment.

A robot systems integrator provided me a hypothetical case where it installs a robotic system on an LHD that runs three shifts a day, each lasting eight hours, and winds up showing enormous cost savings.

Before going further with this case study study, let me pause for a brief moment to alert the reader that there is a "catch" to all of this that I will discuss later that may involve freeing up bottlenecks and the possibility of low cost human labor alternatives. However, this still does not take anything away from this hypothetical example under the right set of circumstances. Furthermore, as I explain elsewhere in this paper, from a very long term viewpoint a company has no choice but to give advanced technology the benefit of the doubt.

Continuing with the hypothetical example, prior to installation, the LHD ran five hours per eight hour shift. The robotic system is designed to extend operations an extra two hours a shift to seven hours, with one hour left over for maintenance.

In this particular scenario, teleoperation does not result in unemployment for any of the three LHD drivers who work each of the three shifts. It simply moves them from the mine to the teleoperation center on the surface where they can each work one shift apiece in a twenty four hour period.

The up front capital for the teleoperation equipment comes to about $2 million. This includes the control room as well as devices on the vehicles. The cost of installation is $400,000.

The extra two hours a shift of LHD movement means six extra hours a day, or 2,160 hours in a 360 day year. This would cause a company to add an extra maintenance person for each additional 2,000 maintenance hours a year. That means $46,000 in added salary, and $50,000 in incremental costs.

The critical question is what additional revenue can be generated by an extra six hours of LHD operation per 24 hour period? The pro forma assumes increased revenue of $1,392 an hour, which is based on certain assumptions involving hypothetical ore yields and commodity prices. All of this yields an Internal Rate of Return (IRR) on a five year pro forma of 68%.




Initial Installation Costs          
........ Installation Costs
........ Capital  
Additional Maintenance Costs
.........Maintenance Costs
.........Maint. Labour
Labour Savings
.........Labour - Savings
.........Additional Revenue
Cash Flow  
.........Net Cash Flow
.........B/E (in months)
cumulative hours

A Business Case for Mine Automation. The application involves a standard block cave mining method. All figures are courtesy of an actual robot systems integrator. Hypothetical numbers assume 3 LHDs with a gain of two hours a shift with three shifts per day. The revenue assumptions are very sensitive to commodity prices, the grading of different types of ore, and other factors. Installation costs and capital are subject to change.


In economics jargon, the pro forma above is a "marginal utility" analysis that examines the revenue impact from an incremental two extra hours of through put per shift. It leaves open the question of how the incremental cost achieved by adding automation compares to the incremental costs of adding more human labor in order to achieve the same two hour per shift increase in throughput.

Instead of buying a teleoperation system, management might also have the option of adding more labor to raise its throughput from 15 hours to 21 hours a day. As one alternative, it might consider going to a four shift system a day in which human LHD operators overlap each other to minimize vehicle downtime. Under conditions of rising commodity prices, adding human labor can also show significant ROI and NPV numbers.

Intellibot, a company that makes robotic floor cleaners for large commercial spaces, has created a chart that shows a typical relationship between incremental human labor costs and increased automation costs. (I also portray this in Part Four of my "I, Robot Investor" series). Typically automation is more expensive in the short run, but becomes cheaper as it gets amortized over the long run over either higher volumes or greater levels of usage. The blue line, marked "Intellibot," starts off at lower volume being more expensive than the brown, green, or red lines that designate human labor. However, as the size covered increases, the "Intellibot" blue line reflects an increasing difference in lower overall costs compared to any of the manual equipment alternatives.










Size of Facility

[source: INtelliBot]


The bottleneck factor: From an operations management perspective, mining is simply a big process flow problem. Using a hard rock gold mine as an example, it starts with the drilling and blasting, and then it all ends when gold is filtered and processed into ingots. All process flows have bottle necks, where by definition the process flow is most constrained. As soon as one bottleneck is freed up, then the next most constrained part of the process flow becomes the new bottleneck.

Sometimes the biggest bottleneck has more to do with commodity prices than the production process. Some extractive industries choose to slow down or stop production if they think they can fetch higher commodity prices later. Some companies such as Silver Standard Resources (SSRI) and Vista Gold (VZG) have gone to the extreme of leaving their reserves in the ground during the last decade of depressed precious metals prices with the intent of moving forward to production only once certain price targets are achieved. Obviously if they ever choose to become operators, they will never even think about robots until commodity prices reach certain levels.

Huge ROI numbers for adding robots typically assume that the robots will free up major bottlenecks. They also assume that mine operators are motivated to maximize throughput.

There are a number of bottlenecks that robots are not yet been designed to overcome. As some examples, they may include frequent maintenance interventions on mining equipment that disrupt production. They may include frequent blasting schedules that force temporary shut downs, or frequent set up times if mining operations are constantly being moved around. In these cases, robots may help reduce the long term costs of certain segments of production, but they are unlikely to increase throughput and show dramatic ROI numbers.

However, even in marginal situations, there are significant advantages to automating. One is safety, whereby mining companies can remove more workers from harm's way. The need to spend money on ventilation and other factors to support humans is reduced. As mentioned, automation may also increase efficiency and lower overall costs over the long run by boosting efficiency.

Ultimately, the pressure to automate never goes away. As mentioned earlier, the Japanese Robotics Association and other sources project accelerating implementation of mobile robotics, as robots steadily deliver increasing performance and ease of use at lower costs. Automation not only involves workers shaping the behavior of robots, but also the reverse. Robots help to reshape a corporate culture to become more responsive to technology and innovation. Since teleoperators manage several robots, mine workers who get involved in teleoperation are effectively promoted into supervisory and management positions.

The quality of the underlying human organization will always remain a major issue. It is hard to add robots to make incremental efficiency improvements if the underlying human organization is generally inefficient and unresponsive even to conventional efficiency improvement methods. It is hard to add "technological load" (increase technological content and complexity to improve output) if for political reasons an organization has become embedded with incompetent people on all levels who have problems dealing with reality in general, not to mention technological change. I discuss technology vs. politics in greater depth in Part Four of my "I, Robot Investor" series. Some American companies have become so heavily politicized as a result of government-mandated social re-engineering programs that they dare not fire any members of "protected classes" and instead feel that they must continually sweep massive waste and incompetence under the rug. It may be easier to build a robotized mining company from scratch than to try to adapt an existing company by eliminating dead wood.

Clearly the totally automated mine of the future will require different types of robots designed to handle every single phase of the production process. In essence, we would be creating a mobile factory assembly line adapted to mining operations. In the early phases of "robotization," it may make sense to use an "ag ant" or "coolie labor" approach with lots of robots performing finite tasks until increasing levels of robot intelligence allow for more versatility. This would help create a continuous mechanical excavation and ore processing system, yielding the ultimate in efficiency.


Currently Placer Dome's "MiniMole" project is still under wraps in the development phase. According to Placer Dome's April 2003 Research and Technology bulletin:

One of the most exciting projects under way is the development of a novel machine that will allow the `surgical' mining of narrow vein and reef-type deposits. This mechanical mining method is designed to improve safety by removing miners from the active rock face in underground mines. It also aims to drastically reduce the amount of development and waste dilution that is experienced with conventional methods.

Dubbed the MiniMole, fabrication of the first prototype of this new machine was completed in December 2002. In early 2003, quarry trials began with the support of several consulting groups, including Cellula Robotics Ltd., AG Associates LLC and International Submarine Engineering Ltd.

Placer Dome's MiniMole prototype undergoing tests.
Photo source: Placer Dome.

This is all Placer Dome is prepared to publicly release at this point. The fact that Placer Dome talks about removing miners from the rock face and about employing robot companies to help build this machine suggests that this could be a teleoperated second generation robot. By second generation, I mean that unlike the LHDs that I have already described, these self-propelled machines are designed purely for autonomous operation or teleoperation and were never designed to seat a human driver.

This is also second generation to the extent that the robot is designed to start drilling into veins and "surgically" keep going without worrying about whether the surrounding environment is supportive of human life. Most of the robotized vehicles I have talked about so far function in areas designed to support people. The fact that Placer Dome has hired a submarine engineering company may suggest that Placer Dome will not be overly concerned if the robot enters areas flooded with water. In fact, Placer Dome signaled its interest in submersible operations even further with its December 7, 2004 announcement regarding the formation of Placer Dome Oceania Ltd. to mine the depths of the Southwest Pacific.

The word "surgical" has some other interesting connotations. This gets back to a theme often seen in robotics that the "end effector," or the part that does actual work, may be more important than other robotic modules. While Placer Dome is not releasing details on its particular boring device, it is helpful to be aware of state-of-the-art developments elsewhere.

Fred C. Delabbio, Ph.D., P.Eng., released in Sept 2003 a PowerPoint presentation on the Hatch WRBA 2003 conference titled "Expanding the limits of mechanical excavation." On page 28 he shows the MiniMole, and describes it as being for "remote mining narrow vein deposits."

On page 29 he shows the Oscillating Disc Cutter (ODC) developed by Terratec and Odyssey. ODC technology was invented by David Sugden in Australia in the late 1990's to dramatically increase the efficiency of sustained rock cutting. According to a "current & emerging rock cutting technology" technical paper:

...This cutting technique is the application of disc force to the rock face in such a way as to propagate a tensile failure of the rock, if this force can be exerted many times per second (High frequency). To attenuate the sympathetic harmonics that are created, a reactive mass surrounds the cutter. The advantage of this system is that the assembly should be able to be mounted on a relatively light support platform (Vehicle).

On page 27 of his Hatch presentation, Dr. Frank Delabbio shows the Sandvik Tamrock ARM 1100 which uses an undercut disc cutting technology. It started as a joint venture in 1999 between Lonmin Platinum and Voest-Alpine, a division of Sandvik Tamrock. Mike Woof had some excellent commentary on this new device in the April 2002 World Mining Equipment article "The Hills Are Alive":

Voest-Alpine's other highly innovative project is for a narrow reef miner, intended to take the place of conventional drill and blast techniques for South Africa's increasingly important platinum mining industry. These mines are currently making a large step forward in their operating methods by turning to much higher levels of mechanization, so as to increase productivity and replace old-fashioned hand drilling techniques. The traditional methods are labor intensive, result in excessive footwall development, headgrade dilution, require ore tramming and also pose safety issues for the miners. Voest-Alpine reckons it can capitalize on the need for new and productive equipment from the platinum miners with a machine that is designed specifically for the application. This utilizes a novel disc cutting method, so there is no need for blasting and Voest-Alpine reckons the concept will provide a very competitive and cost-effective mining solution. The machine is designed for use in narrow reef heights from 850-1,200 mm and with rock hardnesses of 40-210 MPa. The machine should reduce safety risks as it can be remotely operated as well as improving infrastructure utilization and headgrade. However, as the machine is radical and still highly experimental, Voest-Alpine is not able to say much about the project at this stage of its development. The first prototype was test run in Austria prior to shipment and has since begun its initial trials at Lonmin's platinum operations in South Africa, having been commissioned on-site late last year. Although specific details of its cutting method and design have yet to be revealed, WME will be watching closely as this information and results of the machine's performance become available.

The term "surgical" reinforces the robotic theme of redefining the work concept. A robot that can follow veins and reefs does not produce as much waste rock as human operations that have to widen tunnels to allow human access. This increases the profitability of throughput.

Perhaps we can now see a robot "tortoise" win against a human "hare." I would make an analogy here with the Roomba vacuum cleaner created by iRobot corp, which recently sold its millionth model. The Roomba, which runs off a semi random algorithm, takes much longer than a human to vacuum clean a room. But if all a human has to do is "turn it on and walk away" (the Roomba sales slogan), what do he or she care if the Roomba takes longer to clean, particularly since it has a self-docking feature at a recharge station?.

Similarly, if you have a whole army of MiniMoles surgically gong after vein formations, operating day and night, why do you necessarily care that each MiniMole may take a bit longer to excavate rock than humans equipped with drills and blasting equipment?

Last, but not least, this kind of machine may substantially augment the ability of underground mining companies to grow their reserves by the drill bit. This aspect was dramatized in the Dec 4, 2004 Financial Sense Newshour interview by James Puplava with Chris Davie, Director, President and CEO of Queenstake Mining

Chris Davie: [Queenstake's] Jerritt is four underground mines. The perception that we are trying to correct in the market place is that it has a small reserve. Its reserve equates to 2.5 years of production. The answer is that we will have between 2 to 3 years of reserves for the next 10 years. It continues to regenerate its reserves on an annual basis.
James Puplava: Is this is characteristic of underground mines?.
Chris Davie: It is, absolutely. And the classic example is the Homestake mine which was finally shut down in the last two or three years after 135 years of operation. The maximum reserve that Homestake ever had ahead of it was 2 years.

I understand from a Placer Dome source that the MiniMole may have some capability in the future to funnel the ore it drills back to a tramming area through a trailing hose. It's propulsion methods are currently a trade secret. It is not designed to be completely submerged, but rather to function in an aerated environment. Rather than pump ore back to a tramming station, it will more likely climb at an upward diagonal angle and let gravity help slide the ore backwards. Part of the issue here is the ability to create a robotic system that can create a continuous process, which in turn leads us to third generation robotic concepts that I will discuss next.


Dr. Greg Baiden , Chairman and Chief Technical Officer of robotics systems integrator Penguin ASI, holds the Canadian Research Chair in Mine Automation and Robotics. Recently he has been conducting field research on the concept of using lasers underwater to control submersible teleoperated robots. His former employer Inco is considering drilling at its Creighton mine between 5 km to 10 km below the surface. Rock bursts from earth pressures become common below 5 km and typically kill several Canadian miners every year. In addition, drill holes can close up.

One solution is to pressurize mines by flooding them with water and using robo-sub miners. Dr. Baiden is interested in lasers as means to control the subs without using cords that can tangle when several robots work on the same object.

According to "Kinross, Rio Tinto Eye underwater mining technology" the Kinross VP of Technology Services thinks the laser approach might be useful for open pit mines as well. Teleoperation has problems on the surface because it requires high bandwidth, and this can infringe on government-mandated frequency assignments. Lasers might be a way around this.

Water is usually the underground miner's enemy. It must be constantly pumped out. The submersible approach converts it into a vital ally. We no longer have to worry about ventilation, temperature control, noise reduction, structural supports, and other costs necessary to support humans. Water pressure can counteract tremendous earth pressures. The aqueous environment also allows our robo-miners to use buoyancy to efficiently ferry ore both horizontally and vertically. A long tube that connects the mining robot to processing facilities on the surface would also eliminate the need to break bulk between the ore carried by an LHD vehicle on a deep underground horizontal tramming plane and the ore carried in a vertical elevator to the surface. I also understand from a company that supplies submersible robots to the offshore oil and gas industry that compressed air bubbles can help drive ore through a tube to the surface.

Admittedly retrieval robots would be needed to pull seriously malfunctioning and immobilized robots back to dry dock for humans to perform maintenance. However, if the entire underwater production chain can be successfully robotized and supported with a large logistical infrastructure, one might wonder if it might also some day make sense to flood open pit and shallow mines as well. A circular effect might apply here, where aqueous mining may force miners to achieve total robotization, which in turn may encourage creating more aqueous environments to utilize fully robotized systems.

One problem with the laser approach being developed by Dr. Baiden is that it is line-of-sight. Furthermore, mining machines require extraordinary amounts of sustained power. This makes it advantageous to keep them connected to a cord that supplies power, not to mention the teleoperation lines that can go with the power cords.

I believe that this should not necessarily be a one-or-the-other debate. One could use both approaches simultaneously. For example, one could have a "mother ship" sent deep underground that is connected by cord to the surface. The cord could include not only teleoperation and electrical lines , but also hoses for any gases or fuels that might be also used to produce mechanical energy. The mother craft could then act as a docking station for lots of smaller worker bots that are controlled by laser or by extension cords from the mother ship. They could return to the mother ship to get recharged or re-fueled. The worker bots could also serve as laser relay stations with each other. They could also also be produced in many different sizes with many different types of functionality to aid continuous mechanized ore processing.

We can even go a step further. If connecting a submersible sub to a cord becomes a problem, we might consider making the cord itself a robot. This kind of concept is being pursued with "snake bots" at various universities such as the University of Southern California and Carnegie Mellon. Each cord segment would have its own computer chip which would enable it to connect itself with other segments in a long line, or to alternatively detach itself to form various exotic shapes. The snakebot cord would also be able to coil itself, wiggle, spiral, or combine with other cords to form a longer cord.

Last, but not least, in order to achieve optimal efficiency, one of our ultimate goals is to create a continuous robotic production process. With a submersible system, it may become possible to cut out the tramming step altogether, and simply pump ore generated by the drilling robot through a long hose that leads to the surface of the mine, aided by rising compressed air bubbles. The long tube would also carry power, teleoperation, and air/fuel gas resupply lines. The mining process could then begin to have similarities with horizontal drilling methods already being used in the oil and gas industry.


Typically innovation occurs as both a "bottom up" as well as a "top down" process within an organization. In Part Five of my robot series I explain why a focus on simply replacing humans with robots is often the wrong approach from both a business and social perspective.

From a "bottom up" perspective, miners have a lot going for them. They are used to maintaining equipment and improvising under constantly changing conditions. They understand the tasks that need to be performed better than anyone else. They can be an invaluable source for developing practical new robotic system ideas.

Organizations such as F.I.R.S.T. and the ROBOlympics have demonstrated considerable success in stimulating interest in robotics on a grass roots level through robotic competitions for high school students and adults. At RoboNexus, Dean Kamen, the founder of F.I.R.S.T. noted that China and India graduated 3.4 million engineers in the past year, whereas the United States graduated only 62,000. He commented, "The US needs to change its culture so scientists and engineers, not athletes, are heroes. [If it doesn’t] this country will continue to get what it celebrates.”

I think that mining companies should explore creating organizations and incentive plans that encourage their employees to embrace the "robolution." I also think that the robot revolution will offer so many dynamic opportunities for productivity enhancement that it will pay for mining companies to develop their in-house staff presence in this area, so that they can more quickly create advanced automation to suite their own particular needs. To remain competitive, it may be too risky to wait on mining equipment suppliers to second guess their robot needs for them.

From a "top down" perspective, in the "mix and match" modular world of mobile robotics, the same operating system that may enable a robot to clean sludge out of a pipe might also be used for a more glamorous humanoid application. Mining offers a range of challenges from the very simple to the very complex. Solving these problems can advance general robotic technology and be applied across a wide variety of industries. With rising commodity prices, the mining industry may be in a better position than most American industries to help pay for advanced robotic research and development. This may be particularly true in view of the increasing financial stress being experienced by the consumer, American corporations, and the US Government, as documented by the Grandfather Economic Reports, by Boston University professor Laurence Kotlikoff, and also by financial commentators such as Dr. Marc Faber, Jim Rogers, Bill Murphy, and James Puplava. As alluded to earlier, by advancing robotic technology, mining companies might find themselves not only helping themselves, but also serving vital strategic interests.

I applaud Placer Dome for taking the initiative with their MiniMole robot and their new company Placer Dome Oceania Ltd. I hope that it may become de rigueur for other mining companies to get aggressive about pursuing their own second generation robot projects. Maybe some will get really aggressive and start pursuing third generation concepts. This is the kind of initiative and competition we need.

When I attended the Carnegie Mellon Robotics Institute 25th Anniversary Event in October, it featured a new "Robot Hall of Fame" Perhaps mining organizations should also consider creating their own robot Hall of Fame awards some day. The robot future for the mining industry looks bright for those who bravely follow its path.


This report is for research/informational purposes only, and should not be construed as a recommendation of any security. Information contained herein has been compiled from sources believed to be reliable. There is however, no guarantee of its accuracy or completeness.

Bill Fox welcomes phone calls and responses to this article. His web site and most current contact information is at www.amfir.com.



Flag carried by the 3rd Maryland Regiment at the Battle of Cowpens, S. Carolina, 1781

© William Fox. Sometimes William Fox offers viewpoints that are not necessarily his own to provide additional perspectives.